Date of Award

12-2021

Degree Type

Report

Degree Name

Master of Science (MS)

Department

Economics and Finance

Committee Chair(s)

Tyler Brough (Committee Chair), Todd Griffith (Committee Co-Chair)

Committee

Tyler Brough

Committee

Todd Griffith

Committee

Paul Fjeldsted

Abstract

There has been extensive literature written about the efficiency of the stock market. Practitioners and academicians have debated whether investors can exploit publicly available information to generate excess returns. Clearly predicting the stock market’s return with high accuracy has been enormously difficult, but we are interested in contributing to the continuous exploration of the efficiency of the stock market using machine learning techniques. We also want to examine the relationship between our dataset’s macroeconomic indicators and foreign nations’ stock markets with our target feature—the S&P 500. In this paper, we will be using supervised machine learning models, like Linear Regression, Penalized Regression-Elastic Net, Support Vector Regression, Random Forest, and XGBoost models, to predict monthly stock market returns using historical data from 1992 to 2021. Our results show that it is difficult to forecast stock market returns with high accuracy using the monthly SP&500 monthly returns, and that if investors even rely on the high computational power of machine learning techniques to attempt to forecast stock market returns, they will likely end up making a high-risk bet and lose out substantially on their investment. We also report our dataset features’ importance in relation to the U.S. SP&500 generated by our machine learning model.

Share

COinS